Re: FRBRization in LT, was: Personal perspectives on catalog use

From: Jonathan Rochkind <rochkind_at_nyob>
Date: Tue, 17 Feb 2009 14:55:08 -0500
To: NGC4LIB_at_LISTSERV.ND.EDU
Yeah, and if it's that tricky even in a more-or-less 'closed' system 
that you control (it's all hosted on LT and edited only through LT 
software, although it's edited by many people) -- even trickier to solve 
in the kind of distributed system the library shared bib corpus is.

I am not convinced that this is a problem that we need to tackle.  I 
think a schema that represents a work-set group as either "in" or "out" 
is sufficient, it's not neccesary to try to solve that 
existential/ontology problem otherwise.

Of course, in the real world, we don't just have one 
internally-consistent system. It may be (and in fact already is!)  that 
in the OCLC system a certain book belongs to one workset, while in the 
(eg) LT system it belongs to another. And to make matters more 
confusing, even the worksets themselves won't necessarily have a one to 
one correspondence. We can think about how to write software that can 
take this conflicting info from multiple systems, and reconcile it into 
some kind of coherent view for our users. That view certainly could 
reveal the existence of different determinations of work-set-membership, 
if we can figure out clever ways to display this information without 
being overwhelming.

That seems to me like a lot more productive avenue than trying to figure 
out how to make a schema that makes it possible for workset membership 
to be based on more-than-two value logic.

Jonathan

Tim Spalding wrote:
> No, we're binary now too. You can write "Disambiguation notices," and
> these sometime get at some of the differences. But a non-binary system
> would be hard. We've thought about providing links between works with
> a defined number of standard relationships, but also an option to
> write you own—so you can connect Paradise Lost to Frankenstein and His
> Dark Materials and explain why. You could probably vote things up and
> down—something we use in other contexts. I'm not sure that would be a
> good idea, though.
>
> I would say, though, that the RDF triplet subject predicate object—the
> sky has the color blue—miss nuances like "how blue?" "who's looking at
> the sky?" "is that man your friend?" etc.
>
>   
>> I think ultimately ANY model of reality inside software data structures is
>> going to be 'shoehorning reality into boxes'. That's just the nature of the
>> beast.  The map is not the territory. Our task is to make the map
>> sufficiently good and flexible, to make sure we've got the right choice of
>> boxes to shoehorn things into, including setting up the framework for future
>> uses we can't think of now  But there's no way to get reality itself inside
>> a database, it's always only going to be a model.
>>     
>
> Yes, absolutely. But we can represent things that weren't represented
> before. Even the card catalog represented things that didn't make it
> into computers—the age of the card, its wear, the informative but
> amateur correction made by a scholar. Library systems need to add all
> that back in.
>
> Tim
>
>   
Received on Tue Feb 17 2009 - 14:56:32 EST